README.md

metadata

inference: false
language:
  - zh
license: other
model_creator: lmsys
model_link: https://huggingface.co/lmsys/vicuna-33b-v1.3
model_name: vicuna-33b-v1.3
model_type: vicuna
pipeline_tag: text-generation
quantized_by: shaowenchen
tasks:
  - text2text-generation
tags:
  - gguf
  - vicuna
  - chinese

Provided files

Name	Quant method	Size
vicuna-33b-v1.3.Q2_K.gguf	Q2_K	13 GB
vicuna-33b-v1.3.Q3_K.gguf	Q3_K	15 GB
vicuna-33b-v1.3.Q3_K_L.gguf	Q3_K_L	16 GB
vicuna-33b-v1.3.Q3_K_S.gguf	Q3_K_S	13 GB
vicuna-33b-v1.3.Q4_0.gguf	Q4_0	17 GB
vicuna-33b-v1.3.Q4_1.gguf	Q4_1	19 GB
vicuna-33b-v1.3.Q4_K.gguf	Q4_K	18 GB
vicuna-33b-v1.3.Q4_K_S.gguf	Q4_K_S	17 GB
vicuna-33b-v1.3.Q5_0.gguf	Q5_0	21 GB
vicuna-33b-v1.3.Q5_1.gguf	Q5_1	23 GB
vicuna-33b-v1.3.Q5_K.gguf	Q5_K	21 GB
vicuna-33b-v1.3.Q5_K_S.gguf	Q5_K_S	21 GB
vicuna-33b-v1.3.Q6_K.gguf	Q6_K	25 GB
vicuna-33b-v1.3.Q8_0.gguf	Q8_0	32 GB

Usage:

docker run --rm -it -p 8000:8000 -v /path/to/models:/models -e MODEL=/models/gguf-model-name.gguf hubimage/llama-cpp-python:latest

Provided images

Name	Quant method	Compressed Size
shaowenchen/vicuna-33b-v1.3-gguf:Q2_K	Q2_K	12.78 GB
shaowenchen/vicuna-33b-v1.3-gguf:Q3_K	Q3_K	14.81 GB
shaowenchen/vicuna-33b-v1.3-gguf:Q4_K	Q4_K	18.24 GB
shaowenchen/vicuna-33b-v1.3-gguf:Q5_K	Q5_K	21.72 GB
shaowenchen/vicuna-33b-v1.3-gguf:Q6_K	Q6_K	25.05 GB
shaowenchen/vicuna-33b-v1.3-gguf:Q8_0	Q8_0	31.34 GB
shaowenchen/vicuna-33b-v1.3-gguf:full	full	56.07 GB

Usage:

docker run --rm -p 8000:8000 shaowenchen/vicuna-33b-v1.3-gguf:Q2_K

and you can view http://localhost:8000/docs to see the swagger UI.